John G was correct by telling us that air turbulances can act as
reflectors of ultrasonic waves. I verified this today by using an
ordinary hair dryer (CONAIR CORPORATION, model 088, at 1200W).
Here are illustrations colleced when the averging=3 and sampling
rate=10 per second. [Fluctuations in a were reduced by about a
factor of 5, or so, when averaging was changed from 3 to 15.]
Fluctuations in accelerations were less than 0.05 when the hair dryer
was directed away from the detection area (a test for the absence of
an electronic effect from the motor, etc.). Blowing air horizontally
across the detection area produced a lessstrong fluctuations, as below:
The conclusion from this is that the experiment must be repeated by
placing the detector in front of a moving object, not behind it.
It is possible that we were measuring distances from the turbulance
tail behind the ball and that the tail itself was growing with speed.
An obvious source of a systematic error.
I was told that a chicken wire can be placed directly in front of a
detector (to protect it from the ball) without interfering with
distance measurements. I wish somebody else could confirm the
findings. This would explain why motion detectors were "not very
good" in measurements of g. Another argument for using camcorders?